Do Swarm cluster nodes join the Storidge cluster, or they are separate?
The Storidge software runs hyper-converged with the Swarm, i.e. they can be the same nodes. Just as Swarm provides orchestration for containers, Storidge provides orchestration for volumes in the cluster.
Do all Swarm nodes need to be members of the Storidge cluster, or can Storidge nodes be a subset of Swarm?
Not all Swarm manager nodes have to be in the Storidge cluster. The Storidge cluster just needs to have at least one Swarm manager node.
How can I setup a Storidge cluster without running your install script? We bootstrap Swarm nodes already.
The install script handles installation into pre-existing clusters. It first checks for Kubernetes and Docker swarm. If neither is present that it configures a Swarm cluster and launches a Portainer service. If either K8S or Swarm is present, it only setups the CIO cluster.
For pre-existing Swarm clusters, the Storidge software still needs 3 raw block devices per node, i.e. no filesystems or partitions. The install will automatically discover and add any raw devices available to the storage pool.